P-1996. Performance Evaluation of an Artificial-Intelligence-Driven Single-Cell Imaging Platform for Rapid Phenotypic Antimicrobial Susceptibility Testing

Abstract Background Conventional microbiological methods require several days from specimen receipt to organism identification and antimicrobial susceptibility testing (AST), delaying appropriate therapy. We evaluated an early prototype of the PhAST assay, a novel artificial-intelligence (AI)-driven single-cell imaging platform for rapid phenotypic AST. Methods The prototype PhAST assay was developed to provide bacterial identification at the group level defined by the Clinical and Laboratory Standards Institute, and phenotypic AST results were obtained in ninety minutes, directly from a positive blood culture. Susceptibility was determined by AI-based quantitative analysis of single-cell phenotypes from images and videos of antibiotic-exposed and untreated cells (Figure 1). The assay was performed on positive blood culture broth samples flagged by a routine blood culture system obtained from the clinical laboratory. Only samples containing Gram-negative bacteria were tested. We also measured the time to result (TTR) of the assay. Broth microdilution (BMD) was performed in triplicate as the reference method. Results from the commercial automated AST system (VITEK2) were also assessed against BMD. Results A total of 122 Enterobacterales, including 69 (56.6%) Escherichia coli, 37 (30.3%) Klebsiella pneumoniae, and 6 (4.9%) Enterobacter cloacae complex, were identified. Resistance to third-generation cephalosporins was observed in 54 (44.3%) of the isolates, based on BMD results. The prototype PhAST assay showed a mean (± standard deviation) TTR of 117 (± 3) minutes. The assay achieved favorable results, with a categorical agreement (CA) of 95.1%, and very major error (VME), major error (ME), and minor error (mE) rates of 0.2%, 1.1%, and 3.6%, respectively (Table 1). The commercial AST system exhibited CA, ME, and mE rates of 93.9%, 0.4%, and 3.9%, respectively, with a slightly high VME rate of 1.9%. Conclusion The prototype PhAST assay provided acceptable performance for Enterobacterales within two hours of blood culture positivity. This rapid phenotypic AST assay has significant potential to support timely clinical decision-making. Further expansion to additional organisms and antibiotics could enhance its clinical utility. Disclosures Jeong Su Park, M.D. / Ph.D., PhAST Corp.: Advisor/Consultant|PhAST Corp.: Grant/Research Support Kyoung-Ho Song, MD, PhD, PhAST Corp.: Advisor/Consultant|PhAST Corp.: Grant/Research Support

Authors

Institutions

Publication Details

Journal
Open Forum Infectious Diseases
Published
2026-01-01
DOI
https://doi.org/10.1093/ofid/ofaf695.2161
Primary Topic
Bacterial Identification and Susceptibility Testing
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

P-1996. Performance Evaluation of an Artificial-Intelligence-Driven Single-Cell Imaging Platform for Rapid Phenotypic Antimicrobial Susceptibility Testing

H. R. Kim, Hong Bin Kim, Seong Jin Choi, Jeong Su Park et al.
Open Forum Infectious Diseases
Bacterial Identification and Susceptibility Testing
article

P-1996. Performance Evaluation of an Artificial-Intelligence-Driven Single-Cell Imaging Platform for Rapid Phenotypic Antimicrobial Susceptibility Testing

H. R. Kim, Hong Bin Kim, Seong Jin Choi, Jeong Su Park, D. W. Shin, Song Mi Moon, Yunsang Choi, Kyunghwa Lee, Eu Suk Kim
article en

Abstract

Abstract Background Conventional microbiological methods require several days from specimen receipt to organism identification and antimicrobial susceptibility testing (AST), delaying appropriate therapy. We evaluated an early prototype of the PhAST assay, a novel artificial-intelligence (AI)-driven single-cell imaging platform for rapid phenotypic AST. Methods The prototype PhAST assay was developed to provide bacterial identification at the group level defined by the Clinical and Laboratory Standards Institute, and phenotypic AST results were obtained in ninety minutes, directly from a positive blood culture. Susceptibility was determined by AI-based quantitative analysis of single-cell phenotypes from images and videos of antibiotic-exposed and untreated cells (Figure 1). The assay was performed on positive blood culture broth samples flagged by a routine blood culture system obtained from the clinical laboratory. Only samples containing Gram-negative bacteria were tested. We also measured the time to result (TTR) of the assay. Broth microdilution (BMD) was performed in triplicate as the reference method. Results from the commercial automated AST system (VITEK2) were also assessed against BMD. Results A total of 122 Enterobacterales, including 69 (56.6%) Escherichia coli, 37 (30.3%) Klebsiella pneumoniae, and 6 (4.9%) Enterobacter cloacae complex, were identified. Resistance to third-generation cephalosporins was observed in 54 (44.3%) of the isolates, based on BMD results. The prototype PhAST assay showed a mean (± standard deviation) TTR of 117 (± 3) minutes. The assay achieved favorable results, with a categorical agreement (CA) of 95.1%, and very major error (VME), major error (ME), and minor error (mE) rates of 0.2%, 1.1%, and 3.6%, respectively (Table 1). The commercial AST system exhibited CA, ME, and mE rates of 93.9%, 0.4%, and 3.9%, respectively, with a slightly high VME rate of 1.9%. Conclusion The prototype PhAST assay provided acceptable performance for Enterobacterales within two hours of blood culture positivity. This rapid phenotypic AST assay has significant potential to support timely clinical decision-making. Further expansion to additional organisms and antibiotics could enhance its clinical utility. Disclosures Jeong Su Park, M.D. / Ph.D., PhAST Corp.: Advisor/Consultant|PhAST Corp.: Grant/Research Support Kyoung-Ho Song, MD, PhD, PhAST Corp.: Advisor/Consultant|PhAST Corp.: Grant/Research Support

Open Forum Infectious DiseasesVol. 13(Supplement_1)
Seoul National University (KR), Seoul National University Bundang Hospital (KR)
Openalex Percentile: Top 90%
Bacterial Identification and Susceptibility Testing
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.